Today I read a paper I enjoyed and learned from about the physiology of bacterial lag phase. I do recommend the paper - Rolfe et al., J Bac. 2012 - it is very good, but it nonetheless reminded me of something that has bothered me for a long time: I despise the terms “up-regulated” and “down-regulated.” These terms are usually invoked when a gene expression measurement was conducted (RNAseq, RT-PCR, Western blot, maybe proteomics) and some gene(s) level increased or decreased in some condition: “P. aeruginosa PAO1 specifically downregulated the operons involved in siderophore synthesis” to pick an arbitrary example from the abstract of Lei et al. 2020.

On a purely aesthetic level, “up-regualted” and “down-regulated” are cumbersome, inelegant terms. Mouthfuls of marbles. In papers and talks I stumble over them constantly; I can’t believe my fellow scientists think they merit the esteem accorded by dropping the hyphen. On a scientific level, they are misleading to the point of irresponsibility. The premise of “up-regulation” is the “regulation” - that there is a specific regulatory “program” leading to these gene(s) rising or falling in certain conditions. Of course this is often true, or true enough, but the premise, I think, clouds our thinking more than it helps us.

For one, genetic regulatory programs are complex and intertwined: in many cases a single factor (“mechanism”) can’t honestly be blamed for the whole change in expression. Consider one example dear to my mentor Dianne: expression of phenazine biosynthesis genes is affected by quorum sensing and oxygen tension (Sakhtah et al. 2013) as well as phosphate availability (McRose & Newman, 2020), and likely several other environmental and physiological factors. Moreover, while most assays of gene expression are static measurements of single timepoints, gene regulation is a dynamical process: the expressed RNA and protein accumulates, becomes inactivated, degrades, etc. These dynamical processes are usually not measured, and so the notion that a few static measurements capture a change that is “regulated” rather than transient is therefore usually a supposition.

Last and definitely not least, statements about “down-regulation” are usually made without reference to what’s going on in the rest of the cell. You’ll often see sentences like: “Statistical significance was determined with the Benjamini and Hochberg multiple testing correction, and genes that were differentially expressed by more than 2-fold were identified” (Methods of Rolfe et al. 2012). Who says 2-fold (or any other fold) is a good cutoff to decide that something is “regulated”? What if the average gene’s expression level changed by 1.5 fold because growth slowed by 1.5 fold? For an example of this in yeast see Keren et al., MSB 2013 and for a recent example in E. coli, see Mori et al. MSB 2020.

Maybe I’m a stick in the mud (I am) but I don’t think we should use terminology that encodes our suppositions about the nature and mechanism of gene regulation. Moreover, the terminology is mostly extraneous - you can write a sentence that is honest and about the same length, something like “we observed a roughly threefold increase in the level of 3 genes involved in phosphate uptake.” In any case I would need to read you methods to understand how you measured the increase and how you decided that it differed from the null model.